Transaction index
All transactions are on Solana devnet, and each link opens Solana Explorer. Before these docs were published, every signature below was checked against devnet (how).
The MSTRx cross, 23 Sept: the auction with outside participants
Section titled “The MSTRx cross, 23 Sept: the auction with outside participants”Three orders from three wallets, two placed by people other than the builder. Walked through in Verify a clearing price.
| Step | Signature |
|---|---|
Cross (compute_clearing) |
2SKcwefY…K75FpLAU |
| Settlement | 4ws2ByFq…hNGBKY8A |
The community auction, 24 Sept
Section titled “The community auction, 24 Sept”Announced publicly ahead of time, run on devnet on the regular AAPLx and IBMx books. 11 wallets other than the test bot’s placed orders across the two books, and every one got a fill. The bot’s own orders were in both books (Honest limitations). Both prices were set by volume alone, with no tie, so neither the oracle nor the midpoint rule played any part. verify-auction.mjs matches both.
AAPLx. 24 orders from 13 wallets, 10 of them not the bot. Cleared 53.16 shares at $336.97. Auction 7uc3uXqH…YYMivcVn.
| Step | Signature |
|---|---|
| Opened, 12:42:42 UTC | 5iDs9xEK…6NMMfAnW |
| Cross, 13:02:06 UTC | 439jNtYv…wqCHk2wC |
| Settlement 1 of 4 | 3GVouS2V…N9Bv9ckH |
| Settlement 2 of 4 | 44yyhpbt…JwDgfXc8 |
| Settlement 3 of 4 | 3X5M32Ea…7irP2hJY |
| Settlement 4 of 4 | BWTFG3M8…4tB7Go2P |
IBMx. 20 orders from 12 wallets, 9 of them not the bot. Cleared 28.97 shares at $207.27 a share. That is $210.44968 per token × 28.53 tokens on chain, at IBMx’s multiplier of 1.0153. Auction JgkL8nLK…WBe6WnwD.
| Step | Signature |
|---|---|
| Cross, 12:58:33 UTC | 2MvdyHfb…ymrSCk1K |
| Settlement 1 of 3 | 5FffySLM…Fi6iuJew |
| Settlement 2 of 3 | 62e7BqRi…S5mgHLeD |
| Settlement 3 of 3 | 3xyXxwjR…gHQJxko4 |
The Pyth check recorded “stale” for AAPLx and “wrong owner” for IBMx, which has no Pyth account (What Pyth is used for).
The refund-path fix, 24 Sept
Section titled “The refund-path fix, 24 Sept”cancel_and_refund now requires a paused ticker mint, or a refund path already chosen (Honest limitations). Tested with uncross/scripts/refund-guard-test.mjs on the dormant CRCLx fixture, whose mint could be paused without touching a live book.
| Step | Signature |
|---|---|
| Program upgrade (last-deployed slot 503490825) | 3aE7J9JC…LC42YEKY |
The case the path exists for. Auction ChSD9Wr6…ZHK7ZftG.
| Step | Signature |
|---|---|
| Mint paused and auction crossed, one transaction (4 shares at 205) | EvBMugfE…nkSscS9w |
settle_batch while paused: fails, Token-2022 0x43 |
4AzbxeEA…RSdSwjwX |
cancel_and_refund for the buy, sent by a stranger while paused: succeeds |
RKDH7XZM…nmMNoxu9 |
| Mint resumed | 594sz2k5…zbHR1q4X |
cancel_and_refund for the sell after resume, path already chosen: succeeds |
jRM3XCiY…zejUx2Gw |
Refusals. Each was sent with preflight skipped, so it landed on chain as a failed transaction with RefundNotAllowed (6024).
| Case | Signature |
|---|---|
Mid-window (auction 3EhBRJBu…74mqix) |
3B4HuV4K…KioU2vtf |
| Crossed and unsettled, sent by a stranger | 5t5XBj7c…sfwFMq9z |
| Crossed and unsettled, sent by a participant | 5n2tcn6C…aCJxUomA |
Partly settled: one order clear-settled in the same transaction (auction GyjX3r7r…4JqQENG) |
5vfnRZ2H…spQt9vv6 |
A real browser wallet on the live site, 24 Sept
Section titled “A real browser wallet on the live site, 24 Sept”A wallet that had never existed, in a fresh browser, through Wallet Standard. It took a faucet grant and placed two orders through the page.
| Step | Signature |
|---|---|
| Cancel while open (accepted) | 5xuRETXF…FJPT3eNg |
Cancel sent straight to the program once frozen (refused on chain, PastFreezeWindow) |
wzwxgAAi…GyaZ3ikF |
| Settlement of the second order, shown on its receipt | sf9HDGvT…6uQt4Lbf |
The script that drives it is uncross/scripts/order-path-browser.mjs.
The worked example: 247.50, on the upgraded program
Section titled “The worked example: 247.50, on the upgraded program”Auction Dw4UdSGL…ttm8nt8U. See The clearing rule.
| Step | Signature |
|---|---|
initialize_auction |
3bkfFfn9…vRD4EZBJ |
compute_clearing |
4xEomFan…85X3Ve1T |
settle_batch [0..5] |
4bJG1KMZ…1BnU8jCv |
| Program upgrade to the zero-copy layout | 2Q14KghF…MciKRmCW |
The same book, first run (Phase 1)
Section titled “The same book, first run (Phase 1)”Auction Bddkdu6qTgVNiMecNLZrd1yYQdTEYuk5qft7B9Bt43U4, on the original program. This account uses the earlier, smaller layout and does not decode with today’s.
| Step | Signature |
|---|---|
| Program deploy | 3gNrx6wH…pd3y1hcv |
initialize_auction |
4LWcYzZD…kNpSCCJt |
place_order #0 (sell 10 @ 240) |
2HMSToea…DBDrwfQ1 |
place_order #1 (buy 10 @ 250) |
2ZE9hmg1…oJH7U25g |
place_order #2 (sell 5 @ 245) |
32GZvbu4…iZB35AwG |
place_order #3 (buy 8 @ 243) |
4gNatvg7…ggoUVAjs |
place_order #4 (sell 1 @ 300) |
3W7x1fub…v5tzYEQz |
cancel_order #4 (before the freeze) |
5Vk6fZ9J…G5daYvHC |
place_order #5 (sell 1 @ 310) |
Kb2u57eG…qqh6beQ5 |
cancel_order #5 (in the freeze) |
refused, PastFreezeWindow; no transaction landed |
compute_clearing |
324CU6VC…UPeyFL3P |
compute_clearing again (no-op) |
5QzRNSSQ…yYy4kv9c |
settle_batch [0..5] |
2QkuCd6V…khqCifhF |
Pause and split tests
Section titled “Pause and split tests”Listed in full, with what each step showed, in Test: pause mid-auction and Test: split mid-auction.
Settlement at scale
Section titled “Settlement at scale”The 42-wallet stress run is listed in full in Test: 42-wallet settlement.
The earlier 38-order, 38-owner book was cranked to completion by the production keeper (auction 9uBAAY4gFtD9kG9EPYb8oU9wExRdkGRKmejqERk4r74V, cleared at 100.37 × 190):
| Step | Signature |
|---|---|
compute_clearing |
3PqGDbUV…mxuubbVP |
| settle [0–6] | 2zSMaAAc…6CxUtzWh |
| settle [7–13] | 3gsdcCko…D1GxzbCC |
| settle [14–20] | 4GW28JuC…5WjqJ5BK |
| settle [21–27] | wHEm3NBs…CEbE2wGR |
| settle [28–34] | 5rYzSPJ7…msTXVBor |
| settle [35–37] | 5jx9PE94…Bsv7fcn2 |
The web app’s own transaction code
Section titled “The web app’s own transaction code”The functions the UI calls on click were driven end to end against devnet on the IBMx fixture, with local keypairs standing in for a browser wallet (web/scripts/wallet-flow.ts). Sell 3 @ $100, buy 3 @ $110, buy 1 @ $50 then cancelled. Cleared at $105.00 × 3 shares, both vaults zero.
| Step | Signature |
|---|---|
| sell 3 @ $100 | 2uFQVk7B…GEKLApDV |
| buy 3 @ $110 | 2KGpK3rD…Sgf4mSx2 |
| buy 1 @ $50 | 2NAPv3cn…jfR5vgeZ |
| cancel #2 | 2voYdJUa…2B9hZaNY |
compute_clearing |
22AZrNnW…x2k1QUmA |
| settle | 2jCyg2YV…3z28ZeM4 |
Rent recovery, 23 Sept
Section titled “Rent recovery, 23 Sept”120 close_auction transactions, returning 2.1994 SOL: the 118 recovered stranded auctions plus two opened on dormant tickers during testing. Every one is listed, with the auction, the rent returned and who sent it, in docs/rent-recovery-2026-09-23.tsv. See Honest limitations.
| Signature | |
|---|---|
| First close | 5YdN7KdX…ChFUqsW3 |
| Last close | 4r127fgH…LWxUbmCs |
How this list was checked
Section titled “How this list was checked”The docs site’s build check extracts every Explorer transaction link on every page. It asks a devnet RPC for each signature’s status, with transaction history search on, and fails if any is not found. The cancel refused with PastFreezeWindow above is a transaction that landed and failed, which is how a refusal is recorded on chain. It is expected to show an error.
Sources: docs/submission-draft.md (community auction, commit 70ca5d4), docs/phase2.md (refund-path fix, commit 532dcb7), checked against the auction accounts, their order accounts and transactions on devnet; docs/phase1.md, docs/phase2.md, docs/submission-draft.md, README.md, docs/rent-recovery-2026-09-23.tsv.